[— title: “Research Stuff” output: pdf_document: default html_document: default —

Perguntas!

Quais as perguntas que já foram respondidas?

Comparação de jogadores não contaminados com os outros jogadores

Distribuição:

Percentage of non-contaminated players from the total of enemy groups.

nrow(groups[relation.offender=='enemy' & contamination == 0])*100/nrow(groups[relation.offender=='enemy'])
## [1] 66.99673

Percentage of non-contaminated players from the total of groups.

nrow(groups[relation.offender=='enemy' & contamination == 0])*100/nrow(groups) 
## [1] 20.54079

Comparação de médias:

Não contaminados vs. contaminados:

Non-contaminated groups also presents higher average performance \((mean=0.120, sd=0.02)\) than contaminated groups \((mean=0.086,sd=0.03)\).

med.t.test(groups[relation.offender=='enemy' & contamination == 0]$performance,
                     groups[relation.offender!='enemy' | contamination > 0]$performance)
## [1] "mean/sd x"
## [1] 0.1204023
## [1] 0.02679578
## [1] "mean/sd y"
## [1] 0.09016112
## [1] 0.04093299
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 1022.3, df = 3090500, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.03018321 0.03029917
## sample estimates:
##  mean of x  mean of y 
## 0.12040231 0.09016112

Não Contaminados vs. inimigos contaminados:

Non-contaminated groups also present higher average performance \((mean=0.120, sd=0.02)\) than contaminated \((mean=0.107, sd=0.03)\) enemy groups.

med.t.test(groups[relation.offender=='enemy' & contamination == 0]$performance,
                     groups[relation.offender=='enemy' & contamination > 0]$performance)
## [1] "mean/sd x"
## [1] 0.1204023
## [1] 0.02679578
## [1] "mean/sd y"
## [1] 0.1078593
## [1] 0.0342237
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 258.56, df = 1051300, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01244796 0.01263812
## sample estimates:
## mean of x mean of y 
## 0.1204023 0.1078593

Não Contaminados vs. Aliados contaminados:

And higher average performance \((mean=0.120, sd=0.02)\) than ally groups \((mean=0.084, sd=0.03)\) as well.

med.t.test(groups[relation.offender=='enemy' & contamination == 0]$performance,
                     groups[relation.offender=='ally']$performance)
## [1] "mean/sd x"
## [1] 0.1204023
## [1] 0.02679578
## [1] "mean/sd y"
## [1] 0.08435106
## [1] 0.03088467
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 1122.4, df = 3069900, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.0359883 0.0361142
## sample estimates:
##  mean of x  mean of y 
## 0.12040231 0.08435106

Inimigos Contaminados vs. Aliados (performance):

We can also compare contaminated enemy groups and ally groups, and confirm that contaminated enemy groups \((mean=0.107, sd=0.03)\) have higher average peformance than allies\((mean=0.08, sd=0.03)\).

med.t.test(groups[relation.offender=='enemy' & contamination > 0]$performance,
                     groups[relation.offender=='ally']$performance)
## [1] "mean/sd x"
## [1] 0.1078593
## [1] 0.0342237
## [1] "mean/sd y"
## [1] 0.08435106
## [1] 0.03088467
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 490.9, df = 1018900, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.02341435 0.02360207
## sample estimates:
##  mean of x  mean of y 
## 0.10785927 0.08435106
med.t.test(groups[relation.offender=='ally' & topic=='complaints']$performance,
groups[relation.offender=='ally' & topic!='complaints']$performance)
## [1] "mean/sd x"
## [1] 0.07351362
## [1] 0.02574566
## [1] "mean/sd y"
## [1] 0.08697285
## [1] 0.03143763
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -277.03, df = 687200, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01355445 -0.01336401
## sample estimates:
##  mean of x  mean of y 
## 0.07351362 0.08697285

med.t.test(groups[relation offender==‘ally’ & topic==‘complaints’]\(performance, groups[relation.offender=='ally' & topic!='complaints']\)performance)

Inimigos Contaminados vs. Aliados (contaminação):

Finnaly, comparing the contamination index of contaminated enemy groups and ally groups, we see that the contaminated enemy groups \((mean=0.28, sd=0.15)\) still are less contaminated than the allies \((mean=0.30, sd=0.21)\) but with a small difference.

med.t.test(groups[relation.offender=='enemy' & contamination > 0]$contamination,
                     groups[relation.offender=='ally']$contamination)
## [1] "mean/sd x"
## [1] 0.285968
## [1] 0.1556121
## [1] "mean/sd y"
## [1] 0.2999964
## [1] 0.2185822
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -56.476, df = 1549800, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.0145153 -0.0135416
## sample estimates:
## mean of x mean of y 
## 0.2859680 0.2999964

T-test’s \(p\) is \(2.2*10^{16}\), for all these comparisons.

Distribuição dos tópicos(Positive and Negative):

Por tipo de grupo:

Non-Contaminated:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##    4.154204   16.267211    5.452007    2.498494    3.092136    1.526151 
##     tactics tactics.pos 
##   24.843725   42.166072
##      neg      pos 
## 14.06579 85.93421

Contaminated Enemies:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##   12.377604   17.810885    8.224055    2.540251    2.490208    3.250765 
##     tactics tactics.pos 
##   20.719097   32.587133
##      neg      pos 
## 27.06669 72.93331

Ally:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##   15.811375   10.616322   19.521756    5.354951    3.468746    3.163654 
##     tactics tactics.pos 
##   20.808496   21.254699
##     neg     pos 
## 45.4275 54.5725

Offender:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##   14.923031   10.530200   27.097166   12.001722    2.774965    7.824406 
##     tactics tactics.pos 
##   15.997512    8.850999
##      neg      pos 
## 63.61152 36.38848

Por categoria de tópico

Dataset completo:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##   12.708834   12.485591   18.355101    7.018207    3.025940    4.615350 
##     tactics tactics.pos 
##   19.795542   21.995435

Positive Topics:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##     0.00000    23.00365     0.00000     0.00000     0.00000     0.00000 
##     tactics tactics.pos 
##    36.47162    40.52473

Negative Topics:

##   arguments   chit.chat  complaints     insults other.langs   provoking 
##    29.76482     0.00000    42.98871    16.43705     0.00000    10.80942 
##     tactics tactics.pos 
##     0.00000     0.00000

Relacionamento entre topicos/grupos e performance/contaminação:

Positive topics:

Correlações:

Performance:

## [1] "Todos:"
## [1] 0.6228213
## [1] "Não Contaminados:"
## [1] 0.8046749
## [1] "Inimigos Contaminados:"
## [1] 0.8432692
## [1] "Aliados:"
## [1] 0.9680895
## [1] "Ofensor:"
## [1] 0.9546973

Contaminação:

## [1] "Inimigos:"
## [1] -0.6963913
## [1] "Aliados:"
## [1] -0.7837376

Comparisons

## [1] "Full"
## [1] "mean/sd x"
## [1] 0.1057205
## [1] 0.03809381
## [1] "mean/sd y"
## [1] 0.0854443
## [1] 0.03992133
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 637.95, df = 5682400, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.02021393 0.02033852
## sample estimates:
## mean of x mean of y 
## 0.1057205 0.0854443
## [1] "Non-Contaminated"
## [1] "mean/sd x"
## [1] 0.1224147
## [1] 0.02538513
## [1] "mean/sd y"
## [1] 0.1093002
## [1] 0.03142648
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 167.8, df = 218820, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01296134 0.01326770
## sample estimates:
## mean of x mean of y 
## 0.1224147 0.1093002
## [1] "Contaminated Enemies"
## [1] "mean/sd x"
## [1] 0.1134019
## [1] 0.03214273
## [1] "mean/sd y"
## [1] 0.09398463
## [1] 0.03541024
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 198.06, df = 280720, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01922509 0.01960940
## sample estimates:
##  mean of x  mean of y 
## 0.11340188 0.09398463
## [1] "Allies"
## [1] "mean/sd x"
## [1] 0.09042173
## [1] 0.03159107
## [1] "mean/sd y"
## [1] 0.07722164
## [1] 0.0284678
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 301.98, df = 1879000, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01311442 0.01328576
## sample estimates:
##  mean of x  mean of y 
## 0.09042173 0.07722164
## [1] "Offenders"
## [1] "mean/sd x"
## [1] 0.09873421
## [1] 0.05031007
## [1] "mean/sd y"
## [1] 0.08633709
## [1] 0.04514018
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 188.84, df = 1631700, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01226844 0.01252578
## sample estimates:
##  mean of x  mean of y 
## 0.09873421 0.08633709

Tactics-related(tactics + tactics.pos):

Correlações:

Performance:

## [1] "Todos:"
## [1] 0.5928543
## [1] "Não Contaminados:"
## [1] 0.7640866
## [1] "Inimigos Contaminados:"
## [1] 0.8192308
## [1] "Aliados:"
## [1] 0.9513276
## [1] "Ofensor:"
## [1] 0.9363036

Contaminação:

## [1] "Inimigos:"
## [1] -0.701972
## [1] "Aliados:"
## [1] -0.5905795

NhemNhemNhem

## [1] 0.05058955
## [1] 0.6081889
## [1] 0.4261171

Mood-related(tactics.pos + small talk):

Correlações:

Performance:

## [1] "Todos:"
## [1] 0.5935066
## [1] "Não Contaminados:"
## [1] 0.9501938
## [1] "Inimigos Contaminados:"
## [1] 0.95
## [1] "Aliados:"
## [1] 0.9633967
## [1] "Ofensor:"
## [1] 0.6135424

Contamination:

## [1] "Inimigos:"
## [1] -0.6732689
## [1] "Aliados:"
## [1] -0.8073669

Comparações:

Performance:

Non-contaminated:

## [1] "mean/sd x"
## [1] 0.1242648
## [1] 0.02529473
## [1] "mean/sd y"
## [1] 0.1149968
## [1] 0.02785384
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 195.14, df = 1103700, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.009174971 0.009361142
## sample estimates:
## mean of x mean of y 
## 0.1242648 0.1149968

Contaminated enemy:

## [1] "mean/sd x"
## [1] 0.1144905
## [1] 0.03257593
## [1] "mean/sd y"
## [1] 0.1013424
## [1] 0.03450893
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 157.35, df = 641980, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01298425 0.01331180
## sample estimates:
## mean of x mean of y 
## 0.1144905 0.1013424

Ally:

## [1] "mean/sd x"
## [1] 0.09301305
## [1] 0.03316181
## [1] "mean/sd y"
## [1] 0.0802906
## [1] 0.02886421
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 260.46, df = 1081600, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01262671 0.01281818
## sample estimates:
##  mean of x  mean of y 
## 0.09301305 0.08029060

Offender:

## [1] "mean/sd x"
## [1] 0.1007124
## [1] 0.05380595
## [1] "mean/sd y"
## [1] 0.08834091
## [1] 0.04551635
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 145.58, df = 640900, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.01220493 0.01253804
## sample estimates:
##  mean of x  mean of y 
## 0.10071240 0.08834091

Negative Topics:

Correlações:

## [1] "Todos:"
## [1] -0.5358722
## [1] "Não Contaminados:"
## [1] -0.8778429
## [1] "Inimigos Contaminados:"
## [1] -0.916787
## [1] "Aliados:"
## [1] -0.9518999
## [1] "Ofensor:"
## [1] -0.3573428
## [1] "For performance < 0.06:"
## [1] 0.8339901
## [1] "For performance >= 0.06:"
## [1] -0.8853053

Comparisons:

Comparisons

## [1] "Full"
## [1] "mean/sd x"
## [1] 0.0854443
## [1] 0.03992133
## [1] "mean/sd y"
## [1] 0.1057205
## [1] 0.03809381
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -637.95, df = 5682400, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.02033852 -0.02021393
## sample estimates:
## mean of x mean of y 
## 0.0854443 0.1057205
## [1] "Non-Contaminated"
## [1] "mean/sd x"
## [1] 0.1093002
## [1] 0.03142648
## [1] "mean/sd y"
## [1] 0.1224147
## [1] 0.02538513
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -167.8, df = 218820, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01326770 -0.01296134
## sample estimates:
## mean of x mean of y 
## 0.1093002 0.1224147
## [1] "Contaminated Enemies"
## [1] "mean/sd x"
## [1] 0.09398463
## [1] 0.03541024
## [1] "mean/sd y"
## [1] 0.1134019
## [1] 0.03214273
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -198.06, df = 280720, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01960940 -0.01922509
## sample estimates:
##  mean of x  mean of y 
## 0.09398463 0.11340188
## [1] "Allies"
## [1] "mean/sd x"
## [1] 0.07722164
## [1] 0.0284678
## [1] "mean/sd y"
## [1] 0.09042173
## [1] 0.03159107
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -301.98, df = 1879000, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01328576 -0.01311442
## sample estimates:
##  mean of x  mean of y 
## 0.07722164 0.09042173
## [1] "Offenders"
## [1] "mean/sd x"
## [1] 0.08633709
## [1] 0.04514018
## [1] "mean/sd y"
## [1] 0.09873421
## [1] 0.05031007
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -188.84, df = 1631700, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01252578 -0.01226844
## sample estimates:
##  mean of x  mean of y 
## 0.08633709 0.09873421

Arguments:

Correlações:

Performance:

## [1] "Todos:"
## [1] -0.6423828
## [1] "Não Contaminados:"
## [1] -0.4298965
## [1] "Inimigos Contaminados:"
## [1] -0.6259034
## [1] "Aliados:"
## [1] -0.9224592
## [1] "Ofensor:"
## [1] -0.6907605

Contaminação

## [1] "Todos:"
## [1] -0.6423828
## [1] "Não Contaminados:"
## [1] -0.4298965
## [1] "Inimigos Contaminados:"
## [1] -0.6259034
## [1] "Aliados:"
## [1] -0.9224592
## [1] "Ofensor:"
## [1] -0.6907605

Comparisons:

Performance:

Non-contaminated:

## [1] "mean/sd x"
## [1] 0.1135278
## [1] 0.03188396
## [1] "mean/sd y"
## [1] 0.1207108
## [1] 0.02649725
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -51.843, df = 57852, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.007454559 -0.006911430
## sample estimates:
## mean of x mean of y 
## 0.1135278 0.1207108

Contaminated enemy:

## [1] "mean/sd x"
## [1] 0.09761533
## [1] 0.03675039
## [1] "mean/sd y"
## [1] 0.1094313
## [1] 0.03355229
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -85.958, df = 99616, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01208542 -0.01154657
## sample estimates:
##  mean of x  mean of y 
## 0.09761533 0.10943133

Ally:

## [1] "mean/sd x"
## [1] 0.0778424
## [1] 0.02943269
## [1] "mean/sd y"
## [1] 0.08556669
## [1] 0.03098617
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -132.87, df = 448410, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.007838233 -0.007610358
## sample estimates:
##  mean of x  mean of y 
## 0.07784240 0.08556669

Offender:

## [1] "mean/sd x"
## [1] 0.08624908
## [1] 0.04867387
## [1] "mean/sd y"
## [1] 0.09152615
## [1] 0.04723439
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -60.413, df = 488860, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.005448276 -0.005105872
## sample estimates:
##  mean of x  mean of y 
## 0.08624908 0.09152615
##   arguments   chit.chat  complaints     insults other.langs   provoking 
##   0.2976482   0.0000000   0.4298871   0.1643705   0.0000000   0.1080942 
##     tactics tactics.pos 
##   0.0000000   0.0000000

Minimum for non-contaminated and contaminated enemies.

## [1] "For non-contaminated"
## [1] 0.1262249
## [1] 0.4280555
## [1] "For contaminated enemies"
## [1] 0.1311179
## [1] 0.6017316
## [1] "For non-contaminated"
## [1] 0.1262249
## [1] -0.8223408
## [1] "For contaminated enemies"
## [1] 0.1311179
## [1] -0.9351802
## [1] "mean/sd x"
## [1] 0.2871684
## [1] 0.1527032
## [1] "mean/sd y"
## [1] 0.285968
## [1] 0.1556121
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 3.1332, df = 369350, p-value = 0.001729
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.0004495086 0.0019513567
## sample estimates:
## mean of x mean of y 
## 0.2871684 0.2859680
## [1] "mean/sd x"
## [1] 0.05502653
## [1] 0.01003389
## [1] "mean/sd y"
## [1] 0.14408
## [1] 0.008394818
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -5950, df = 1481900, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.08908280 -0.08902413
## sample estimates:
##  mean of x  mean of y 
## 0.05502653 0.14408000

Complaints:

Offender’s peak:

## [1] 0.0559267
## [1] 0.8739402
## [1] -0.9259598

Correlações:

## [1] "Todos:"
## [1] -0.6048952
## [1] "Não Contaminados:"
## [1] -0.8843088
## [1] "Inimigos Contaminados:"
## [1] -0.9066414
## [1] "Aliados:"
## [1] -0.9591255
## [1] "Ofensor:"
## [1] -0.5048056
## [1] "Inimigos:"
## [1] 0.6241439
## [1] "Aliados:"
## [1] 0.7716947

Comparisons:

Non-contaminated:

## [1] "mean/sd x"
## [1] 0.1041836
## [1] 0.03166442
## [1] "mean/sd y"
## [1] 0.1213482
## [1] 0.02616682
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -142.26, df = 77320, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01740105 -0.01692806
## sample estimates:
## mean of x mean of y 
## 0.1041836 0.1213482

Contaminated enemies:

## [1] "mean/sd x"
## [1] 0.08755252
## [1] 0.03373041
## [1] "mean/sd y"
## [1] 0.1097983
## [1] 0.03362745
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -145.6, df = 62913, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.02254523 -0.02194632
## sample estimates:
##  mean of x  mean of y 
## 0.08755252 0.10979830

Allies:

## [1] "mean/sd x"
## [1] 0.07351362
## [1] 0.02574566
## [1] "mean/sd y"
## [1] 0.08697285
## [1] 0.03143763
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -277.03, df = 687200, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01355445 -0.01336401
## sample estimates:
##  mean of x  mean of y 
## 0.07351362 0.08697285

Offenders:

## [1] "mean/sd x"
## [1] 0.08290308
## [1] 0.04166307
## [1] "mean/sd y"
## [1] 0.09365105
## [1] 0.04916283
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -169.73, df = 1375900, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01087208 -0.01062386
## sample estimates:
##  mean of x  mean of y 
## 0.08290308 0.09365105

Comparisons in negative topics:

Insults:

####Correlações: Performance:

## [1] "Todos:"
## [1] -0.3951109
## [1] "Não Contaminados:"
## [1] -0.8234762
## [1] "Inimigos Contaminados:"
## [1] -0.825512
## [1] "Aliados:"
## [1] -0.4388466
## [1] "Ofensor:"
## [1] 0.3087258

Contamination:

## [1] "Inimigos:"
## [1] 0.001536159
## [1] "Aliados:"
## [1] 0.1302296

Offender’s peak

## [1] 0.0910003
## [1] 0.8134174
## [1] -0.8080494

Comparisons:

Non-contaminated:

## [1] "mean/sd x"
## [1] 0.1080204
## [1] 0.02999186
## [1] "mean/sd y"
## [1] 0.1207299
## [1] 0.02661796
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -75.988, df = 34139, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01303736 -0.01238170
## sample estimates:
## mean of x mean of y 
## 0.1080204 0.1207299

Contaminated enemies:

## [1] "mean/sd x"
## [1] 0.09311682
## [1] 0.03321781
## [1] "mean/sd y"
## [1] 0.1083559
## [1] 0.03412522
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -57.951, df = 17309, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01575451 -0.01472365
## sample estimates:
##  mean of x  mean of y 
## 0.09311682 0.10835590

Allies:

## [1] "mean/sd x"
## [1] 0.07997617
## [1] 0.02802309
## [1] "mean/sd y"
## [1] 0.08459258
## [1] 0.03100976
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -51.595, df = 119890, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.004791779 -0.004441045
## sample estimates:
##  mean of x  mean of y 
## 0.07997617 0.08459258

Offenders:

## [1] "mean/sd x"
## [1] 0.08290308
## [1] 0.04166307
## [1] "mean/sd y"
## [1] 0.09365105
## [1] 0.04916283
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -169.73, df = 1375900, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01087208 -0.01062386
## sample estimates:
##  mean of x  mean of y 
## 0.08290308 0.09365105

Taunts:

Correlações:

Performance:

## [1] "Todos:"
## [1] -0.2348258
## [1] "Não Contaminados:"
## [1] -0.4600085
## [1] "Inimigos Contaminados:"
## [1] -0.8124259
## [1] "Aliados:"
## [1] 0.5774775
## [1] "Ofensor:"
## [1] 0.3243221

Contaminação:

## [1] "Inimigos:"
## [1] 0.7040565
## [1] "Aliados:"
## [1] -0.4560991

Comparisons:

Non-contaminated:

## [1] "mean/sd x"
## [1] 0.1181662
## [1] 0.02785825
## [1] "mean/sd y"
## [1] 0.1204472
## [1] 0.0267625
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -11.508, df = 20613, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.002669483 -0.001892476
## sample estimates:
## mean of x mean of y 
## 0.1181662 0.1204472

Contaminated enemy:

## [1] "mean/sd x"
## [1] 0.09711102
## [1] 0.0336455
## [1] "mean/sd y"
## [1] 0.1083336
## [1] 0.03414479
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -47.501, df = 22457, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01168568 -0.01075951
## sample estimates:
##  mean of x  mean of y 
## 0.09711102 0.10833361

Ally:

## [1] "mean/sd x"
## [1] 0.0923376
## [1] 0.03400452
## [1] "mean/sd y"
## [1] 0.08408427
## [1] 0.03073199
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 59.629, df = 65317, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.007982049 0.008524621
## sample estimates:
##  mean of x  mean of y 
## 0.09233760 0.08408427

Offender:

## [1] "mean/sd x"
## [1] 0.09487084
## [1] 0.05008511
## [1] "mean/sd y"
## [1] 0.09038789
## [1] 0.04724566
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 37.59, df = 219330, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.004249206 0.004716695
## sample estimates:
##  mean of x  mean of y 
## 0.09487084 0.09038789
##   arguments   chit.chat  complaints     insults other.langs   provoking 
##   0.2976482   0.0000000   0.4298871   0.1643705   0.0000000   0.1080942 
##     tactics tactics.pos 
##   0.0000000   0.0000000

Conclusions(Positive):

Complaints & Insults : Offenders with low performance tends to complain more, and complain about his own team. Complaints:

#summary(groups$performance)
#summary(offender.topics.perf[performance > 0.094]$complaints)
#summary(offender.topics.perf[performance <= 0.094]$complaints)
summary(matches$offender.team)
##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
##   0.0000   0.3607   0.7729   0.9696   1.3460 674.1000
summary(matches[offender.performance > 0.094 | offender.groups!='complaints']$offender.team)
##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
##   0.0000   0.3510   0.7412   0.9419   1.2980 674.1000
summary(matches[offender.performance <= 0.094 & offender.groups=='complaints']$offender.team)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.5273  0.9681  1.1560  1.5670 34.4100

Enemy teams taunts more on high contamination and low performance as a response to the opposing team taunts.

Comparison betweeen the mean contamination of each topic: General: Positive Topics:

    'Positive:'
## [1] "Positive:"
    mean(groups[topic.2=='pos']$contamination,na.rm=TRUE)
## [1] 0.1573973
    'Tactics:'
## [1] "Tactics:"
    mean(groups[topic.2=='pos' & (topic=='tactics')]$contamination,na.rm=TRUE)
## [1] 0.1868681
    'Tactics.pos:'
## [1] "Tactics.pos:"
    mean(groups[topic.2=='pos' & (topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.1375642
    'Small Talk:'
## [1] "Small Talk:"
    mean(groups[topic.2=='pos' & (topic=='chit.chat')]$contamination,na.rm=TRUE)
## [1] 0.1533754
    'Tactics.all:'
## [1] "Tactics.all:"
    mean(groups[topic.2=='pos' & (topic=='tactics' | topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.1584509
    'Mood:'
## [1] "Mood:"
    mean(groups[topic.2=='pos' & (topic=='chit.chat' | topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.1425052

Positive Topics [Ally]:

    'Positive:'
## [1] "Positive:"
    mean(groups[relation.offender=='ally' & topic.2=='pos']$contamination,na.rm=TRUE)
## [1] 0.2718455
    'Tactics:'
## [1] "Tactics:"
    mean(groups[relation.offender=='ally' & topic.2=='pos' & (topic=='tactics')]$contamination,na.rm=TRUE)
## [1] 0.3099444
    'Tactics.pos:'
## [1] "Tactics.pos:"
    mean(groups[relation.offender=='ally' & topic.2=='pos' & (topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.2527472
    'Small Talk:'
## [1] "Small Talk:"
    mean(groups[relation.offender=='ally' & topic.2=='pos' & (topic=='chit.chat')]$contamination,na.rm=TRUE)
## [1] 0.2354061
    'Tactics.all:'
## [1] "Tactics.all:"
    mean(groups[relation.offender=='ally' & topic.2=='pos' & (topic=='tactics' | topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.2810425
    'Mood:'
## [1] "Mood:"
    mean(groups[relation.offender=='ally' & topic.2=='pos' & (topic=='chit.chat' | topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.2469709

Positive Topics[Non Contaminated Enemies]:

    'Positive:'
## [1] "Positive:"
    mean(groups[relation.offender=='enemy' & topic.2=='pos']$contamination,na.rm=TRUE)
## [1] 0.08132393
    'Tactics:'
## [1] "Tactics:"
    mean(groups[relation.offender=='enemy' & topic.2=='pos' & (topic=='tactics')]$contamination,na.rm=TRUE)
## [1] 0.07779355
    'Tactics.pos:'
## [1] "Tactics.pos:"
    mean(groups[relation.offender=='enemy' & topic.2=='pos' & (topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.07479239
    'Small Talk:'
## [1] "Small Talk:"
    mean(groups[relation.offender=='enemy' & topic.2=='pos' & (topic=='chit.chat')]$contamination,na.rm=TRUE)
## [1] 0.1014537
    'Tactics.all:'
## [1] "Tactics.all:"
    mean(groups[relation.offender=='enemy' & topic.2=='pos' & (topic=='tactics' | topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.0759202
    'Mood:'
## [1] "Mood:"
    mean(groups[relation.offender=='enemy' & topic.2=='pos' & (topic=='chit.chat' | topic=='tactics.pos')]$contamination,na.rm=TRUE)
## [1] 0.08281015

Negative Topics

    'Negative:'
## [1] "Negative:"
    mean(groups[topic.2=='neg']$contamination,na.rm=TRUE)
## [1] 0.2819183

Relacionar contaminação com a presença do ofensor no ‘all chat’

med.t.test(matches[offender.all == 0]$enemy.contamination, matches[offender.all > 0]$enemy.contamination)
## [1] "mean/sd x"
## [1] 0.05000988
## [1] 0.1158038
## [1] "mean/sd y"
## [1] 0.1083691
## [1] 0.171027
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = -266.14, df = 1166800, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.05878901 -0.05792944
## sample estimates:
##  mean of x  mean of y 
## 0.05000988 0.10836911
med.t.test(matches[offender.all == 0]$ally.contamination, matches[offender.all > 0]$ally.contamination)
## [1] "mean/sd x"
## [1] 0.3180026
## [1] 0.195484
## [1] "mean/sd y"
## [1] 0.2943188
## [1] 0.2250766
## 
##  Welch Two Sample t-test
## 
## data:  x and not_x
## t = 69.803, df = 897030, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.02301881 0.02434882
## sample estimates:
## mean of x mean of y 
## 0.3180026 0.2943188

Non-contaminated groups dominate the 25% high performance scores.

    x = summary(groups$performance)
    full = nrow(groups[performance >= x[5]])
    part = nrow(groups[performance >= x[5] & relation.offender=='enemy' & contamination == 0])
    print("Non-contaminated")
## [1] "Non-contaminated"
    print(part/full)
## [1] 0.4128436
    part = nrow(groups[performance >= x[5] & relation.offender=='enemy' & contamination > 0])
    print("Contaminated enemies")
## [1] "Contaminated enemies"
    print(part/full)
## [1] 0.1562736
    part = nrow(groups[performance >= x[5] & relation.offender=='ally'])
    print("Allies")
## [1] "Allies"
    print(part/full)
## [1] 0.1550557
    print("Offenders")
## [1] "Offenders"
    part = nrow(groups[performance >= x[5] & relation.offender=='offender'])
    print(part/full)
## [1] 0.275827

Alvos do ofensor em certos tópicos negativos

Presença do ofensor no ‘all’ chat aumenta contaminação no time inimigo

summary(matches[offender.all==0]$enemy.contamination)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.00000 0.00000 0.00000 0.05001 0.00000 1.00000
summary(matches[offender.all >0]$enemy.contamination)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.0000  0.0000  0.1084  0.2000  1.0000
part=nrow(matches[offender.all==0])
full=nrow(matches)
part/full
## [1] 0.2397275

Insults

Podemos ver que a média da atividade do ofensor no canal ‘all’ quando este predominantemente insulta \((mean=0.20)\) é menor do que a média de atividade na predominância de outros tópicos \((mean=0.32)\). Isso mostra que o ofensor dá menos atenção ao grupo inimigo em casos de insulto. Não temos confirmação de que o ofensor foca seus ataques no grupo aliado através da atividade no chat. Contudo, analisando as palavras utilizadas pelo ofensor e a relação entre a quantidade de insultos realizadas pelo ofensor e a contaminação, fica claro que o time aliado é consideravelmente afetado por estes insultos.

x <- matches[offender.groups=='insults']$offender.all
not.x <- matches[offender.groups!='insults']$offender.all
t.test(x,not.x)
## 
##  Welch Two Sample t-test
## 
## data:  x and not.x
## t = -115.85, df = 432660, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.1237800 -0.1196616
## sample estimates:
## mean of x mean of y 
## 0.1993550 0.3210758
x <- matches[offender.groups=='insults']$offender.team
not.x <- matches[offender.groups!='insults']$offender.team
t.test(x,not.x)
## 
##  Welch Two Sample t-test
## 
## data:  x and not.x
## t = -58.656, df = 372260, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.1136733 -0.1063222
## sample estimates:
## mean of x mean of y 
## 0.8958369 1.0058346

Taunts

A média da atividade do ofensor no canal ‘all’ quando este predominantemente provoca \((mean=0.45)\), é significativamente superior \((p < 2.2*10^{16})\) a média da atividade do ofensor no ‘all’ quando outros tópicos são predominantes \((mean=0.29)\). isso mostra que em casos de provocação, o ofensor foca no time inimigo em detrimento do time aliado, que mostra atividade significativamente mais baixa \((p < 2.2*10^{16})\) por parte do ofensor em casos onde provocação é predominante \((mean=0.68)\) do que em outros casos \((mean=1.01)\).

x <- matches[offender.groups=='provoking']$offender.all
not.x <- matches[offender.groups!='provoking']$offender.all
t.test(x,not.x)
## 
##  Welch Two Sample t-test
## 
## data:  x and not.x
## t = 35.775, df = 140500, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.1526535 0.1703497
## sample estimates:
## mean of x mean of y 
## 0.4559313 0.2944297
x <- matches[offender.groups=='provoking']$offender.team
not.x <- matches[offender.groups!='provoking']$offender.team
t.test(x,not.x)
## 
##  Welch Two Sample t-test
## 
## data:  x and not.x
## t = -53.937, df = 141710, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.3486237 -0.3241753
## sample estimates:
## mean of x mean of y 
## 0.6801412 1.0165407

Podemos replicar esse experimento para os tópicos negativos restantes:

Complaints

Diferença estatisticamente significativa em ambos, logo, consideramos que houve um ‘empate técnico’. Contudo, a diferença real no ‘team’ é bem significativa \(mean=1.11\) para grupos com complaints, \(mean=0.95\) para grupos sem complaints. Isso significa que ofensores que reclamam alvejam o time aliado mais do que o normal.

x <- matches[offender.groups=='complaints']$offender.all
not.x <- matches[offender.groups!='complaints']$offender.all
t.test(x,not.x)
## 
##  Welch Two Sample t-test
## 
## data:  x and not.x
## t = 6.5847, df = 1575600, p-value = 4.558e-11
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.004174234 0.007712285
## sample estimates:
## mean of x mean of y 
## 0.3103061 0.3043628
x <- matches[offender.groups=='complaints']$offender.team
not.x <- matches[offender.groups!='complaints']$offender.team
t.test(x,not.x)
## 
##  Welch Two Sample t-test
## 
## data:  x and not.x
## t = 106.71, df = 1313700, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.1608507 0.1668698
## sample estimates:
## mean of x mean of y 
## 1.1088832 0.9450229
# plt <- plot.topic.perf('neg') + labs(x='Group Performance',y='Groups w/ Negative Topics (%)')
# save.plot('negative_perf.png',plt,w=6)
# plt <- plot.topic.perf('arguments') + labs(x='Group Performance',y='Groups w/ Argument Topics (%)')
# save.plot('arguments_perf.png',plt,w=6)
# plt <- plot.topic.perf('complaints') + labs(x='Group Performance',y='Groups w/ Complaint Topics (%)')
# save.plot('complaints_perf.png',plt,w=6)
# plt <- plot.topic.perf('insults') + labs(x='Group Performance',y='Groups w/ Insult Topics (%)')
# save.plot('insults_perf.png',plt,w=6)
# plt <- plot.topic.perf('provoking') + labs(x='Group Performance',y='Groups w/ Taunts Topics (%)')
# save.plot('taunts_perf.png',plt,w=6)
# plt <- plot.topic.cont('neg') + labs(x="Group Contamination", y='Groups w/ Negative Topics (%)')
# save.plot('negative_cont.png',plt)
# plt <- plot.topic.cont('arguments') + labs(x="Group Contamination", y='Groups w/ Argument Topics (%)')
# save.plot('arguments_cont.png',plt)
# plt <- plot.topic.cont('complaints') + labs(x="Group Contamination", y='Groups w/ Complaint Topics (%)')
# save.plot('complaints_cont.png',plt)
# plt <- plot.topic.cont('insults') + labs(x="Group Contamination", y='Groups w/ Insult Topics (%)')
# save.plot('insults_cont.png',plt)
# plt <- plot.topic.cont('provoking') + labs(x="Group Contamination", y='Groups w/ Taunt Topics (%)')
# save.plot('provoking_cont.png',plt)

Tabela de Outcomes(Passar para um arquivo .R eventualmente)

Adicionando ally e enemy targets a matches(Passar para o groups.R eventualmente)

Insults and Taunts - Some negative topics does not affect its own team.

summary(matches[ally.groups=='provoking' & ally.chat.all > 0 & ally.chat.team == 0]$enemy.contamination)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.2000  0.4000  0.3844  0.6000  1.0000
summary(matches[ally.groups=='insults' & ally.chat.all > 0 & ally.chat.team == 0]$enemy.contamination)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.0000  0.2000  0.2927  0.5000  1.0000
summary(matches[enemy.groups=='insults' & enemy.chat.all > 0 & enemy.chat.team == 0]$ally.contamination)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.0000  0.0000  0.1435  0.2500  1.0000
summary(matches[enemy.groups=='insults' & !(enemy.chat.all > 0 & enemy.chat.team == 0)]$enemy.contamination)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.00000 0.00000 0.00000 0.08177 0.20000 1.00000

Offender Topics

Negative Topics

###Correlations Performance:

offender.correlations.perf(column)
## [1] "Não Contaminados:"
## [1] 0.7885939
## [1] "Inimigos Contaminados:"
## [1] 0.8625
## [1] "Aliados:"
## [1] -0.9147327

Contamination:

offender.correlations.cont(column)
## [1] "Inimigos:"
## [1] -0.4042116
## [1] "Aliados:"
## [1] 0.6735054

Complaints

###Correlations Performance:

offender.correlations.perf(column)
## [1] "Não Contaminados:"
## [1] 0.8152262
## [1] "Inimigos Contaminados:"
## [1] 0.8958885
## [1] "Aliados:"
## [1] -0.9231925

Contamination:

offender.correlations.cont(column)
## [1] "Inimigos:"
## [1] -0.6210242
## [1] "Aliados:"
## [1] 0.6936982

Arguments

###Correlations Performance:

offender.correlations.perf(column)
## [1] "Não Contaminados:"
## [1] 0.5768569
## [1] "Inimigos Contaminados:"
## [1] 0.7022608
## [1] "Aliados:"
## [1] -0.4806165

Contamination:

offender.correlations.cont(column)
## [1] "Inimigos:"
## [1] 0.5244392
## [1] "Aliados:"
## [1] 0.3898706

Insults

###Correlations Performance:

offender.correlations.perf(column)
## [1] "Não Contaminados:"
## [1] -0.5121617
## [1] "Inimigos Contaminados:"
## [1] 0.404235
## [1] "Aliados:"
## [1] -0.2365162

Contamination:

offender.correlations.cont(column)
## [1] "Inimigos:"
## [1] -0.6239175
## [1] "Aliados:"
## [1] 0.4408136

Taunts

Correlations

Performance:

offender.correlations.perf(column)
## [1] "Não Contaminados:"
## [1] -0.4894697
## [1] "Inimigos Contaminados:"
## [1] -0.684642
## [1] "Aliados:"
## [1] 0.7340415

Contamination:

offender.correlations.cont(column)
## [1] "Inimigos:"
## [1] 0.7096013
## [1] "Aliados:"
## [1] -0.34537

KDA X Performance

KDA:

Central 50%:

prop.table(summary(players[kda > x[2] & kda < x[5] & outcome != 'Leave']$outcome))
##     Leave      Loss       Win 
## 0.0000000 0.5095367 0.4904633

Lower 25%:

prop.table(summary(players[kda <= x[2] & outcome != 'Leave']$outcome))
##      Leave       Loss        Win 
## 0.00000000 0.90857458 0.09142542

Upper 25%:

prop.table(summary(players[kda >= x[5] & outcome != 'Leave']$outcome))
##      Leave       Loss        Win 
## 0.00000000 0.08091766 0.91908234

Means: KDA full summary:

summary(players$kda)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   0.000   1.083   2.000   2.720   3.400  88.000

Winners kda summary:

summary(players[outcome=='Win']$kda)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   0.000   2.200   3.250   4.076   5.000  88.000

Losers kda summary:

summary(players[outcome=='Loss']$kda)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   0.000   0.700   1.200   1.401   1.857  66.000

Performance:

Central 50%:

prop.table(summary(players[performance > x[2] & performance < x[5] & outcome != 'Leave']$outcome))
##     Leave      Loss       Win 
## 0.0000000 0.4930341 0.5069659

Lower 25%:

prop.table(summary(players[performance <= x[2] & outcome != 'Leave']$outcome))
##      Leave       Loss        Win 
## 0.00000000 0.93510478 0.06489522

Upper 25%:

prop.table(summary(players[performance >= x[5] & outcome != 'Leave']$outcome))
##      Leave       Loss        Win 
## 0.00000000 0.08053301 0.91946699

Means: Performance full summary:

summary(players$performance)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.01368 0.06456 0.09013 0.10000 0.12340 0.61000

Winners performance summary:

summary(players[outcome=='Win']$performance)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.01751 0.09574 0.11940 0.12980 0.15210 0.61000

Losers performance summary:

summary(players[outcome=='Loss']$performance)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.01442 0.05271 0.06720 0.07140 0.08522 0.49810

Comparison

Group performance:

Central 50%:

prop.table(summary(groups[performance > x[2] & performance < x[5] & outcome != 'Leave' & relation.offender != 'offender']$outcome))
##      Loss       Win 
## 0.4989203 0.5010797

Lower 25%:

prop.table(summary(groups[relation.offender != 'offender' & performance <= x[2] & outcome != 'Leave']$outcome))
##        Loss         Win 
## 0.997567401 0.002432599

Upper 25%:

prop.table(summary(groups[relation.offender != 'offender' & performance >= x[5] & outcome != 'Leave']$outcome))
##        Loss         Win 
## 0.003679112 0.996320888

Means: Performance full summary:

summary(groups[relation.offender != 'offender']$performance)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.01994 0.06965 0.10010 0.10030 0.13080 0.20290

Winners performance summary:

summary(groups[relation.offender != 'offender' & outcome=='Win']$performance)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0325  0.1168  0.1307  0.1294  0.1432  0.2029

Losers performance summary:

summary(groups[relation.offender != 'offender' & outcome=='Loss']$performance)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## 0.01994 0.05752 0.06971 0.07122 0.08336 0.17820

Plots